fuzzy completions #308
fuzzy completions #308
Conversation
- configurable to switch back to the previous strict match (REPL-based) completions
|
examples: |
aviatesk
force-pushed
the
avi/fuzzycompletions
branch
from
faaaab2 to
9d2487a
Compare
aviatesk
force-pushed
the
avi/fuzzycompletions
branch
from
9d2487a to
e2021ba
Compare
Codecov Report
@@ Coverage Diff @@
## master #308 +/- ##
==========================================
- Coverage 54.92% 54.88% -0.05%
==========================================
Files 42 42
Lines 2933 2602 -331
==========================================
- Hits 1611 1428 -183
+ Misses 1322 1174 -148
Continue to review full report at Codecov.
|
- reflect fuzzy completion adapter addition - revert back to one-liner statement for minor visibility improvement
|
Alright, I updated precompile statements. It would be nice if I could have a feedback on this :) |
| ) | ||
| end | ||
|
|
||
| function fuzzycompletionadapter( |
There was a problem hiding this comment.
This seems fairly similar to the replcompletionadapter -- maybe we could have an is_fuzzy kwarg and do something like
repl_provider = is_fuzzy ? FuzzyCompletions : REPLCompletions
repl_provider.completions(...)
plus have an if block for the different filter methods below?
There was a problem hiding this comment.
hm, then the code after repl_provider used would be really type-instable (I confirmed that Julia 1.4 can't infer the type of repl_provider.completions, for example)
I haven't taken any actual benchmark, but I would like to endure the verbose code for performance here.
There was a problem hiding this comment.
Hm fair enough then. I don't think the dynamic dispatch would make much of a difference, but let's just keep this as-is then.
There was a problem hiding this comment.
I don't think the dynamic dispatch would make much of a difference
yeah maybe. I may be too nervous about dynamic dispatches..
I will refactor them in the future when I get confidence that dynamic dispatches here don't matter.
should work once JuliaRegistries/General#11930 gets merged.
After that I will refactor the test cases a bit and update precompile stmts.